“Quick to learn and hard to kill.” – @TimParsa
SAFETY
-
Variation and Selection: Evolutionary Principles in AI Development
By
–
Variation (“but”) and selection (“therefore”).
-
ML Model Develops Own Simulation Understanding Through Probing
By
–
Using an ML technique called “probing,” they looked inside the model’s “thought process” as it generated new solutions. After training on >1 million random puzzles, they found that the model spontaneously developed its own conception of the underlying simulation, despite never
-
Hallucinations Leaderboard: Measuring LLM Reliability and Accuracy
By
–
If you want to learn more hallucinations in LLM, I recommend the hallucinations leaderboard (cc @clefourrier
): https://
huggingface.co/spaces/halluci
nations-leaderboard/leaderboard
… It also comes with a nice paper from Hong et al. https://
arxiv.org/abs/2404.05904 -
Claude Family Models Excel with Strong Alignment Safety Measures
By
–
The Claude family from @AnthropicAI is really good at it: Claude 3.5 Sonnet, 3 Opus & Haiku all succeed. It feels like a heavy alignment stage with refusals is part of the explanation.
-
AI Model Hallucinations: Size Doesn’t Guarantee Factual Accuracy
By
–
The "Indigo Sock Game" doesn't exist but most models will hallucinate it for you. Factuality hallucinations are fascinating because they often behave in unexpected ways. You'd think bigger models would perform better, but that's not necessarily true.
-
Criminalizing AI-Generated Harm Through Stronger Regulation
By
–
Even if AI generation leads to harm, let us criminalize the harm – it is immaterial that it was AI-generated. Deep fakes, dangerous chemicals, etc, are harmful no matter how they are generated. @FTC has taken a wise stance on this. Why not strengthen our existing laws to prevent
-
Agent Accuracy Degradation: Impact of Precision Steps
By
–
Well said. Not even 30 steps. With 95% accuracy, the agent's accuracy drops to 69% after just 7 steps, making it unusable in practice. However, with 99.9% accuracy and 30 steps, the agent's accuracy remains over 97%.
-
AI and Humans: Mutual Hallucinations and Misconceptions
By
–
AI hallucinates about humans, and humans hallucinate about AI.